Breakout Summary Report

Breakout Description

Goal. Several measurement needs and gaps for studying clouds and precipitation have been identified by the 2020 ARM Decadal Vision and by the community input collected through the ARM Cloud and Precipitation Measurements and Science Group (CPMSG). To follow up on these needs, the goal of the breakout session is to:

• inform the community about the progress made for some specific needs;


• discuss priorities that can be advanced through collaborative efforts in the ARM/ASR community; and


• discuss priorities that particularly need ideas and feedback from the broader community to form both short-term and long-term strategies.

Main Discussion

Structure of the session. To meet the goals, CPMSG selected the following needs for discussions:

• (30 mins) Need for measuring liquid water path (LWP) in the presence of precipitation:
  
  
  
  • Maria Cadeddu provided a brief review of the status and progress made in LWP retrievals in drizzling conditions using microwave radiometer (MWR) measurements. These retrievals, however, only work for conditions when precipitation does not reach the ground. For conditions in the presence of surface precipitation, a few possible solutions were suggested, including 1) site-specific analysis to understand the site-dependent behavior of precipitation systems and to refine the retrieval method, 2) the use of instrument synergy to fill the missing data when surface precipitation is present, 3) possible trade off accuracy-convergence for physical retrievals, and 4) off-zenith view and instrument hardware design that can be difficult, and no results guaranteed.
  
  
  • Roj Marchand provided his view on past efforts and mentioned a dual polarization rain radiometer for measuring LWP for thick clouds and rain in the presence of surface precipitation.
  
  
  • Some other potential approaches were discussed, including dual-frequency, 3D cloud tomography, mounting microwave radiometer with scanning radar, and exploring the scanning capability and maximizing the synergy between lidar, radar, and MWR.
  
  
  • Open discussions mainly focused on clarifications that aimed to understand why MWR measurements cannot be used in the presence of surface precipitation, and why the past efforts on spinning mirrors to shed out standing water that sits on the MWR’s radome didn’t work. While the standing water on the radome seemed to be the main issue for MWR retrievals to fail, Sergey Matrosov mentioned that it is not an issue for radar retrieval, because radar retrievals rely on the attenuation difference between radar gates and thus are not affected by the standing water (see Matrosov et al., 2009, https://doi.org/10.1175/2009JAMC2106.1)
  
  
  
  

• (30 mins) Need for cloud droplet number concentration retrievals:
  
  
  
  • Damao Zhang reported recent evaluation results of ARM VAP NDROP against the ACE-ENA in situ cloud probe measurements. Other lidar-based and radar-based retrievals were also made and compared. The results showed that NDROP has generally a relatively wide range of cloud droplet number concentrations compared to other retrievals, and the mean of NDROP retrievals appears to be larger than that derived from in situ data. It was concluded that lidar-based retrievals worked well and can be one of the ARM VAPs. Regarding the performance of NDROP, it was mentioned that caution should be exercised in evaluations, especially for optically thin clouds in which NDROP has difficulty performing well due to the uncertainty in MWR-retrieved LWP. The desire of having retrievals and intercomparisons involving other sites (in addition to ENA) and longer periods at ENA (in addition to ACE-ENA periods) is noted.
  
  
  • CPMSG proposed to coordinate community efforts to write a review paper on this topic. Specifically, there is a need to better identify the accuracy and uncertainty required in various applications such as climatology statistics, model evaluation, aerosol-cloud interactions, and cloud-precipitation process studies.
  
  
  • There were a lot of discussions on the scope of the review paper and the approach that we might be able to take to handle the various assumptions in the retrieval methods. To better understand and quantify the retrieval uncertainty, the use of synthetic observations that capitalizes Large-Eddy Simulation output and instrument simulator was suggested. ASR past QUICR (quantification of uncertainty in cloud retrievals) efforts were mentioned. It was also mentioned that retrievals from various methods can be put all together on a unified grid as a combined product for users. Additionally, retrieval methods typically work for certain types of clouds, and running them routinely does not always yield valid results, highlighting the importance of cloud regime classification.
  
  
  
  

• (40 mins) In-cloud velocity:
  
  
  
  • Christopher Williams provided a brief summary of the previous CPMSG breakout session last year, highlighting that there is an equal desire to estimate air motion in warm clouds and ice-containing clouds and that the barriers come from the difficulty of retrieval algorithm development and radar data calibration and quality control.
  
  
  • Discussions focus on the pathways to address data quality control and retrieval algorithm development.
  
  
  • The pros and cons of pathways to obtain QC data are summarized below. The hand vote indicated near equal interest in all three pathways.
    
    
    
    • Open source: The pros are less work for ARM and more control of processing for ASR PIs; the cons are some PIs do not want to relinquish their intellectual property, and ARM is shifting QC responsibility to ASR PIs.
    
    
    • PI products: The con is that PIs need support to contribute PI products.
    
    
    • ARM-curated products: ARM has control of products but requires money and codes.
    
    
    
    
  
  
  
  

• Regarding the retrieval algorithm development, there is an interest in developing community codes to retrieve in-cloud velocities in different regimes.

Key Findings

There is no lack of effort and progress in the past few years in retrieving cloud droplet number concentration and LWP for cloud types of non-precipitating and precipitating in the absence of surface precipitation. It appears important to provide retrievals with proper uncertainty estimates and provide retrievals that are more widely available for AMF deployments, fixed sites (mainly ENA and SGP), and the upcoming LASSO-ENA cases. It is also important to help users to understand the underlying assumptions used in the retrieval products and the impact of assumptions, enabling users to apply appropriate retrievals for their specific studies.

Considering the input about the possible review paper and the maturity of retrieval methods, a possible way to move forward is outlined in Action Items (below) and will be discussed among CPMSG members and potentially the wider community before providing ARM with a formal recommendation.

Issues

N/A

Needs

N/A

Decisions

N/A

Future Plans

N/A

Action Items

Action 1: Establish a unified and robust scheme or products for cloud type classification. Since existing retrieval methods use measurements from various passive and active instruments, some may work particularly well for certain cloud types, and less well for a different cloud condition. To provide high-quality routine cloud products, the first step is to apply retrieval methods for suitable cloud types only. Running retrieval algorithms routinely without carefully filtering invalid retrievals can lead to inappropriate evaluation and analyses.

Action 2: Build a case library for ARM and the wider community to test retrieval methods. Using synthetic observations generated from model output and instrument simulator to test retrieval methods is not a new concept, but it is not a universal practice in the retrieval community. Even if such a practice is performed during the algorithm development, different simulation cases may be used, making it difficult to translate retrieval errors from one method to another. Considering that the concept of case libraries is extremely successful in the modeling community and radiation community for intercomparison activities, this could be an effective starting point for us to coordinate retrieval efforts and establish a benchmark for the community. To make this work, there are two critical components: model output and instrument simulator. Both have existing material for us to start with and move forward. 

Action 3: Perform systematic quantification of retrieval uncertainty. This is the common theme and need across all three subtopics discussed in the breakout session. For some advanced retrieval methods, uncertainty due to input observables and imperfectness of simulators is appropriately incorporated and thus the retrievals are associated with proper uncertainty estimates. For some methods, the method used in QUICR could be adapted. However, all of these do not consider structural errors, i.e., the errors in assumptions. It has been suggested to run retrievals with different assumptions to provide uncertainty estimates. It is worth noting that while users have focused on the accuracy and uncertainty in retrievals of cloud droplet number concentration and LWP, they just want some values for air motion retrievals. Even though there is large uncertainty in air motion, it is still invaluable for the ARM/ASR community at the current stage.  

Action 4: Define uncertainty requirements in cloud microphysical property retrievals. This needs to be defined by the literature and/or data analyses designed to address this particular question. An immediate pathway is unclear. It may be a good item to collect input in a targeted workshop. 

Action 5: Showcase results of cloud microphysical properties at ENA and write up as “intercomparison activity” in BAMS. There are interests in participating in this activity. CPMSG Chair will initiate informal chats and meetings to collect more input, form the group, and identify the scope of the paper. It is possible to combine Action 4 and Action 5 together to emphasize the scientific reasoning for such intercomparison activities and the goal of ARM measurements.

Action 6: Continue collecting input on developing the capability of measuring LWP in the presence of surface precipitation. We will continue collecting new ideas on this aspect and inform ARM about the feasibility assessment.

Action 7: Define input data sets and scope for air-motion retrievals. Two possible input data sets include synthesized instrument measurements derived from model output and instrument simulators (see Action 2) and measurements from a few selected ARM cloud events (i.e., Epochs). These data sets will define the cloud regime and the scope of the air motion retrievals (e.g., warm non-precipitating clouds, drizzling clouds, etc.). The efficacy of published air motion retrieval algorithms will be quantified using these data sets through an intercomparison activity.